scripts/seed_hse_pyqs.py

from __future__ import annotations
import sys
from pathlib import Path

# Add backend directory to sys.path
backend_dir = Path(__file__).resolve().parent.parent
sys.path.append(str(backend_dir))

from app.core.database import SessionLocal
from app.models.previous_paper import PreviousPaper
from app.models.previous_question import PreviousQuestion
from app.models.user import User

def seed_default_hse_papers(db, user_id: str) -> None:
    # 5 years of Physics, Chemistry, Biology, Mathematics
    papers_data = [
        {
            "title": "HSE Physics March 2024",
            "subject": "Physics",
            "year": 2024,
            "questions": [
                {"question_number": "1", "question_text": "State Lenz's Law of electromagnetic induction.", "marks": 1, "chapter": "Electromagnetic Induction", "topic": "Lenz's Law", "question_type": "definition", "difficulty": "easy", "keywords": ["Lenz", "induction", "direction"]},
                {"question_number": "2", "question_text": "Derive an expression for the induced EMF in a moving conductor in a uniform magnetic field.", "marks": 4, "chapter": "Electromagnetic Induction", "topic": "Motional EMF", "question_type": "long", "difficulty": "medium", "keywords": ["motional EMF", "EMF derivation"]},
                {"question_number": "3", "question_text": "Write any two properties of electromagnetic waves.", "marks": 2, "chapter": "Electromagnetic Waves", "topic": "EM Wave Properties", "question_type": "short", "difficulty": "easy", "keywords": ["properties", "transverse"]}
            ]
        },
        {
            "title": "HSE Chemistry March 2024",
            "subject": "Chemistry",
            "year": 2024,
            "questions": [
                {"question_number": "1", "question_text": "What is the unit of rate constant for a first-order reaction?", "marks": 1, "chapter": "Chemical Kinetics", "topic": "First Order Reaction", "question_type": "mcq", "difficulty": "easy", "keywords": ["rate constant", "unit"]},
                {"question_number": "2", "question_text": "Explain the SN2 mechanism with a suitable example.", "marks": 4, "chapter": "Haloalkanes and Haloarenes", "topic": "Nucleophilic Substitution", "question_type": "long", "difficulty": "medium", "keywords": ["SN2", "mechanism", "substitution"]}
            ]
        },
        {
            "title": "HSE Biology March 2024",
            "subject": "Biology",
            "year": 2024,
            "questions": [
                {"question_number": "1", "question_text": "Draw a neat labelled diagram of the female gametophyte of angiosperms.", "marks": 3, "chapter": "Sexual Reproduction in Flowering Plants", "topic": "Embryo Sac Diagram", "question_type": "diagram", "difficulty": "medium", "keywords": ["diagram", "embryo sac", "female gametophyte"]},
                {"question_number": "2", "question_text": "Explain the process of double fertilization in flowering plants.", "marks": 4, "chapter": "Sexual Reproduction in Flowering Plants", "topic": "Double Fertilization", "question_type": "long", "difficulty": "medium", "keywords": ["fertilization", "double fertilization", "syngamy"]}
            ]
        },
        {
            "title": "HSE Mathematics March 2024",
            "subject": "Mathematics",
            "year": 2024,
            "questions": [
                {"question_number": "1", "question_text": "Find the derivative of sin(x^2) with respect to x.", "marks": 2, "chapter": "Continuity and Differentiability", "topic": "Chain Rule", "question_type": "numerical", "difficulty": "easy", "keywords": ["derivative", "chain rule"]}
            ]
        },
        {
            "title": "HSE Physics March 2023",
            "subject": "Physics",
            "year": 2023,
            "questions": [
                {"question_number": "1", "question_text": "Define self-inductance of a coil and write its SI unit.", "marks": 2, "chapter": "Electromagnetic Induction", "topic": "Self Inductance", "question_type": "definition", "difficulty": "easy", "keywords": ["self-inductance", "Henry"]}
            ]
        }
    ]

    for p in papers_data:
        # Check if already seeded to prevent duplication
        existing = db.query(PreviousPaper).filter(
            PreviousPaper.user_id == user_id,
            PreviousPaper.title == p["title"]
        ).first()
        if existing:
            continue

        paper = PreviousPaper(
            user_id=user_id,
            title=p["title"],
            subject=p["subject"],
            syllabus="Kerala HSE",
            year=p["year"],
            file_name=f"{p['title'].replace(' ', '_').lower()}.pdf",
            file_type="application/pdf",
            file_path=f"uploads/hse/{p['title'].replace(' ', '_').lower()}.pdf",
            status="ready",
            extracted_text=f"This is the pre-seeded official textbook grounding for {p['title']}."
        )
        db.add(paper)
        db.commit()
        db.refresh(paper)
        
        saved_questions = []
        for q in p["questions"]:
            question = PreviousQuestion(
                previous_paper_id=paper.id,
                question_number=q["question_number"],
                question_text=q["question_text"],
                marks=q["marks"],
                subject=p["subject"],
                syllabus="Kerala HSE",
                year=p["year"],
                chapter=q["chapter"],
                topic=q["topic"],
                question_type=q["question_type"],
                difficulty=q["difficulty"],
                keywords_json=q["keywords"]
            )
            db.add(question)
            saved_questions.append(question)
            
        db.commit()
        paper.extracted_questions_json = [
            {
                "question_number": q.question_number,
                "question_text": q.question_text,
                "marks": q.marks,
                "chapter": q.chapter,
                "topic": q.topic,
                "question_type": q.question_type,
                "difficulty": q.difficulty,
                "keywords": q.keywords_json,
            }
            for q in saved_questions
        ]
        db.add(paper)
        db.commit()

def seed_all_users() -> None:
    db = SessionLocal()
    try:
        users = db.query(User).all()
        print(f"Seeding {len(users)} users with default Kerala HSE papers...")
        for user in users:
            seed_default_hse_papers(db, user.id)
        print("Seeding completed successfully!")
    finally:
        db.close()

if __name__ == "__main__":
    seed_all_users()